ScholarAI is an AI-powered research assistant built to help users understand and interact with academic or technical content more efficiently. It allows users to upload documents, ask questions, and receive context-aware answers generated using modern language models.
The project was initially developed during a hackathon and later extended to improve structure, usability, and overall performance.
Working with long research papers or technical documents can be time-consuming. ScholarAI addresses this by combining document processing with retrieval-augmented generation (RAG), enabling users to extract relevant insights quickly without manually scanning entire documents.
- Upload and process PDF or text documents
- Ask questions based on uploaded content
- Generate concise summaries of complex material
- Retrieve contextually relevant information using embeddings
- Interactive chat interface for continuous queries
Frontend
- React
- Tailwind CSS
Backend
- FastAPI
- Python
AI / ML Components
- Large Language Models (Mistral / OpenAI)
- Retrieval-Augmented Generation (RAG)
- FAISS for vector similarity search
- Text chunking and embedding pipelines
- A user uploads a document (PDF or text)
- The backend processes the document and splits it into smaller chunks
- Each chunk is converted into vector embeddings
- Embeddings are stored and indexed using FAISS
- When a query is made, relevant chunks are retrieved
- The language model generates a response using this context
This approach ensures that responses are grounded in the actual document rather than being generic or hallucinated.
cd backend
python -m venv .venv
.venv\Scripts\activate # Windows
pip install -r requirements.txt
uvicorn app.main:app --reload